Blog post

Symposium Notes – Day Three Features Data Assembly

By Merv Adrian | October 24, 2016 | 0 Comments

SecurityOSSopen sourceMagic QuadrantIndustry trendsGartnerDBMSdata lakeData IntegrationApache HadoopData and Analytics Strategies

With 24 meetings under my belt from the first two days at Orlando Symposium, Wednesday’s 13 (and a presentation) didn’t look quite as daunting. It began well, with enough time for a muffin and some tea at 730 AM in the analyst workroom near to the cubicle I’d spend the day in. Then I launched right in to a couple of predictive analytics discussions.

You might ask whether that’s a bit off my area of coverage. Well, yes, a bit – I’m really a DBMS guy, and have also been doing plenty of work around Hadoop as well for the last couple of years. But Data and Analytics go well together – that’s our team’s name and the title of our Summits next year, after all. So in true Symposium fashion, I pitched in. Five of Wednesday’s inquiries began with or significantly featured predictive analytics and machine learning, and there was a stronger echo of the interest in conversational, AI-based front ends that I first heard on Tuesday.

But assembling the data, especially from packaged applications, was a bigger theme, especially combining data from packaged apps with other existing and new content – and few (none of those I spoke to) seem to want the app vendor to be the one to own the combination process or the tools, it seems. Interesting dilemma for those vendors, all of whom seem to be investing aggressively in inbuilt analytics, and some in data lakes, which don’t combine the data but create a pile of Legos for the programmers/analysts to assemble.

My actual Hadoop count for the day was low – only two conversations, one from a skeptic yet to do a first project. Despite Hadoop’s earliest promoted role as the mechanism for extraction and subsequent assembly, it came up rarely – even for the analyst (me) most closely identified with Hadoop among the analysts at Symposium.

“Challenges” questions were spread out, and mirrored the issues identified in Nick Heudecker and Jim Hare’s  2016 Big Data survey: skills, security were once again topics and data assembly was identified (as “integrating multiple data sources”) in the top 6. A significant minor theme in my 1on1 conversations was culture and organization; one delegate even told me that “agile is a problem” because it bypasses important governance and rationalization issues.

Other frequent memes popped up: getting to the cloud, getting out of oppressive legacy DBMS vendor stacks, and exploring new DBMS types. A few vendor briefings, of course, and a presentation of the DBMS Magic Quadrants rounded out the day, and left me eager for a little downtime. The US Presidential debate that evening gave me all the “down” I needed. At least that will be over soon.

Comments are closed